HP 3PAR Storage Lineup

For 2015, the flash technology has significantly strengthened its position. According to analysts, the market for All-Flash-systems has grown many times, while the number of sold arrays using only mechanical disks is constantly decreasing.


')
If earlier the purchase of SSD required IT justification before the business, now, often, the justification requires the purchase of "regular" drives.
Building an All-Flash Datacenter is becoming a reality and requires more than just flash; it requires building a new infrastructure that is fundamentally optimized for performance (I / O, response time, throughput). 16Gb FibreChannel becomes mainstream, 32Gb FC is planned to appear soon. It is quite possible that high-speed loseless Ethernet will become even more active in the future for accessing data, or maybe we will see active development of other technologies, for example, NVMe over Fabrics, we'll see.

A lot of interesting things happened in the Flash technologies themselves. 3D NAND is actively entering the market, providing low storage cost and high performance.

The market for storage systems using flash has evolved in several ways.
Manufacturers of traditional arrays went the simplest way - they added new technologies to outdated architectures. And they received neither ultra-low response time, nor high performance, nor technologies for reducing storage costs.

Small startups, in turn, began to develop from scratch, offering the market a system with a fairly high performance and low response time. Some of them have proposed storage cost reduction technologies, for example, on-the-fly deduplication.

By the way, in my opinion, technologies for reducing the stored data volume, such as thin technologies or deduplication, should be implemented only on the fly, inline. After all, if you use deferred processing (post-process), you might get an interesting situation:
Imagine that there is 2TB of free space on the storage system. The customer recorded 1TB of data in one day, overnight deduplication turned them into 500GB. Fine!

The next day, another 1TB was recorded, deduplication worked well overnight, 2: 1. On the third day the situation repeated. What got? 3TB logical data, which occupy only 1.5TB on the disk. Sumptuously! Saved expensive disk resource, reduced TCO.

But suddenly the evil virus or the shaking hands of the administrator destroyed all the data on this logical volume. Catastrophe? Not! After all, the administrator still knows his business and he has a recent backup. And here the most interesting begins. How much was physically free space for this task? 2TB. How much is in bakape? 3TB. Can we order 3TB to 1.5TB on the fly? Not! And this is almost a disaster.

That's why I claim that the technologies for reducing the stored data volume should be implemented only on the fly.

The resulting high-speed systems, unfortunately, turned out to be without the functionality necessary for an enterprise level array. Without high fault tolerance, without scaling, without replication to build a disaster-proof solution, without QoS, without integration with applications. Most without the ability to granularly increase the volume, some even without the ability to add volume without losing data. Well, to the stability of the work there were questions.

These systems turned out to be separate islands in data centers, without integration with the rest of the ecosystem.

Some manufacturers of traditional arrays stated that their current architecture is optimal for flash, but as a result, they still bought a startup with a specialized AFA system. Over time, most of the startups were bought by manufacturers of traditional arrays, who realized that their existing technologies were outdated for the new era, but full integration with their existing products had not yet happened.

3PAR, whose internal architecture was originally designed to best work with data and optimize I / O with the help of a specialized coprocessor, was in a special position. And the emergence of new high-speed carriers did not come as a surprise, because the technology was already ready. Yes, additional optimization was carried out taking into account the best for the SSD block size (Adaptive reads / writes), technologies were developed to obtain more usable space with SSD without reducing the lifetime and performance (Adaptive Sparing), the possibilities for using SSDs themselves were expanded (Adaptive Flash Cache, inline deduplication).

But I didn’t have to change anything fundamentally, and 3PAR reached record-breaking performance with a response time of 0.2 ms, providing the highest level of reliability and enterprise-level services.

3PAR - is a complete line, which includes the system of middle, senior level and All-Flash-system. Unified OS, optimized for Flash, unified management, possibility of federation up to 4 arrays of different levels to ensure data mobility.

Evidence of the correctness of the chosen approach 3PAR are numerous awards. According to IDC CY15 Q3, WW Disk Tracker 3PAR ranked first in sales in the Midrange SAN array segment, is one of the leaders among All-Flash-Arrays. Particularly interesting is the rating of Gartner Critical Capabilities for General-Purpose, Midrange Storage Arrays. In it, the leading industrial analyst evaluates the storage rating in various categories: OLTP databases, consolidation, server consolidation and VDI, analytics, clouds. 3PAR 8000 is a leader in all individual categories and ranks first in the general category, overall use case, which perfectly reflects the ability of 3PAR to consolidate all types of heterogeneous services of an organization.

In 2015, the line was updated, the 8000th and 20,000th series were presented.



The All-Flash model 3PAR 20850 took the first place in the SPC-2 industrial test, which demonstrates the capabilities of storage bandwidth, opening up a new approach to building systems oriented to the highest streaming performance. Previously it was believed that flash-technology is not the best way for tasks that require high streaming bandwidth - analytics, video on demand. But 3PAR 20850 proved that the use of flash in this segment is justified.

- Record result in SPC-2 - 62,844.45 MBPS
- The first all-flash solution in the SPC-2 top-10 price / performance (6th place), $ 19.93 / SPC-2 MBPS
- Two times less than the occupied place in the data center compared to the nearest competitor result, significantly lower power consumption and heat dissipation.

3PAR storage systems regularly ranked first in the SPC-1 test, which focused on determining the efficiency and performance of storage systems when working with traditional tasks of a modern enterprise — transactional databases, virtualization, and postal systems. We expect soon the publication of a record result and a new generation.

The first places in the SPC-1 and in the SPC-2 3PAR (although not obtained at different times) indicate the architectural advantages of the 3PAR, designed for heterogeneous loads. The ability to simultaneously provide, with minimal mutual influence, the highest performance for services with a random access profile with a small block size and services that require high bandwidth and use a large block size, makes 3PAR systems an advantageous offer for data centers built using flash technologies.

I would like to highlight the major changes in the new generation.

- A new generation of multi-core Intel IvyBridge CPUs (recall that in 3PAR, the central processors operate only with metadata, commands, all data flow goes through a specialized coprocessor, ASIC chip).

- New ASIC Gen5 , providing all the work with the data:
- RAID calculation
- communication between controllers and fast data transfer between them
—Thin Provisioning with compression of zeros and SSD Deduplication “on the fly”, a lot of improvements have been made in the new generation related to deduplication
- Optimization for work with heterogeneous load
- New ! End-to-end data integrity monitoring according to the T10 PI (DiF) industrial standard, which provides control from the moment the data enters the storage area network until it is placed on disk. Customer data will always be under additional control and protection from possible previously not detected errors of SAN switches, disks. This greatly increases the reliability of storage, taking into account the increasing possibility of undetected errors with increasing volumes of transmitted and stored data.

- Naturally, more cache and larger volumes on SSD, which can be used under Adaptive Flash Cache
- Significantly higher performance in random access operations
- Capacity is several times higher than previous generation
- 16Gb FC frontend, reduces delays and optimal use of modern SAN infrastructure
- 12GB SAS for connecting disk shelves, with support for optical cabling, which allows the expansion shelves to be spread up to a distance of 25 meters (in the 8000th) and 100 meters (in the 2000th series) , for some customers this is very important
- A lot of improvements in iSCSI- stack (including DCB support to reduce delays and increase productivity), which will allow optimal use of new systems where the customer does not want to use FibreChannel SAN

The HPE 3PAR Adaptive Architecture allows Adaptive Flash Cache, deduplicated volumes, regular volumes, as well as data volumes involved in automatic multi-level storage on the same SSD-drives. This allows you to simplify administration, increase the efficiency of using SSD, save even more and get the most out of SSD.

For those who are still thinking about using SSD drives, within the 3PAR storage system we offer free Flash Advisor toolkit, which allows you to evaluate how effective Adaptive Flash Cache will be, what deduplication ratio for existing data will provide Thin Deduplication, even if it isn’t yet SSD.

It is worth noting that the functionality of Adaptive Flash Cache, Thin Deduplication, Thin Provisioning with zero compression is included in the base OS without the need for additional licenses.

To dispel customer doubts about the use of SSDs, HPE provides a standard unconditional guarantee for a period of five years for SSDs (including the end of the resource for recording and rewriting, Wear-out). At the same time, thanks to proprietary 3PAR technologies, we provide our SSDs with a larger volume than the competitors - 480 GB, 920 GB, 1.92 TB and 3.84 TB without restrictions when used for recording. We offer the opportunity to use a variety of SSD technology options - SLC, eMLC, cMLC, 3D NAND, depending on customer requirements.

In the middle of February, a new 3PAR array of the 20,000th series was announced - HPE 3PAR StoreServ 20840 Converged Flash Array . This is a data storage system similar to the optimized All-Flash system 3PAR 20850, but with the possibility of using mechanical disks for those customers who wish to use a hybrid approach. The system has a record performance of over 3.2 million IOPS with a response time of less than 1 ms.

HPE 3PAR 20840 combines high performance with low response time, six nines reliability with the wide functionality required in a modern enterprise:

- QoS with the unique ability to provide SLA at the same time in terms of performance and response time for business-critical applications. And the minimum you can specify a target of 0.5 ms!

- Scalability up to 15 PB (!) Of the addressed usable space

- Synchronous / Periodic / Asynchronous replication (Remote Copy) with support for 3 data centers, between any models of the 3PAR family.

- Hardware deduplication on SSD using ASIC

- Data encryption with integration with external key servers, compliance with the FIPS 140-2 standard with disk access control

- Availability at Six Nines

- Deep integration with leading applications, application integration with snapshot functionality

- VMware vMSC certification and building distributed storage also for other OS

- High density placement

- 5 years unconditional guarantee for SSD, turn on replacement in case of wear-out

We told only about a small part of the benefits and functionality that HPE 3PAR StoreServ possesses, for more information, contact the distributor - MUK, or the HPE representative.